docker.md

Docker Setup

ask.UOS uses Docker Compose for orchestrating services in production deployments.

Container Architecture

graph TB
    subgraph DockerCompose[Docker Compose]
        Caddy[Caddy]
        Streamlit[app_streamlit]
        Worker[app_worker x N]
        Crawl[Crawl4ai]
        Redis[Redis]
    end
    Caddy --> Streamlit
    Caddy --> Worker
    Worker --> Crawl
    Worker --> Redis
    Streamlit --> Redis

Loading

app_streamlit and app_worker are the Streamlit UI and FastAPI backend as separate services (see docker-compose.prod.example.yml); the dev docker-compose.yml still runs both in one app container. app_worker runs with multiple replicas, load-balanced by Caddy (optional; see Caddy.example). Crawl4ai is the crawl4ai service (web scraping). Redis is the redis service (cache and sessions). RAGFlow/Infinity (or Milvus, if configured) run separately — they are not services in this app's compose file.

Service Definitions

• app_streamlit: Streamlit UI, single replica
• app_worker: FastAPI backend, multiple replicas (deploy.replicas) behind the reverse proxy
• caddy: Optional reverse proxy, round-robin load balancing + health checks across app_worker replicas
• crawl4ai: Web scraping service used by the web search tool
• redis: Caching and sessions

Setup Steps

• cp docs/docker-compose.prod.example.yml ./docker-compose.yml and, if using Caddy, cp docs/Caddy.example ./Caddyfile — edit both for your deployment (image tag, replica count, domain)
• cp src/backend_config_example.yaml src/backend_config.yaml and cp prompts_example prompts -r
• Build and start services with Docker Compose

---

Next: Configuration →